Jeff M.
Many drum VSTs have multi outs - separate outs for kick, snare (top & bottom), overheads, room, toms, etc.
Each will typically have different fx added to them - comps, eq, verb "wetness".
 

Exac-a-lackly!

cparmerlee

SuperG
Drums - can never get them just right in a single stereo track. That's plenty of reason to split them out into multiple audio tracks.

Can you please explain a little more about that?  Do you send them to two different synths?

One or more MIDI tracks go to a single synth that has many audio outputs.

Jeff M.
Many drum VSTs have multi outs -
Each will typically have different fx added to them

OK.  Makes sense.  Thanks.
 
This gets me wondering, maybe an imponderable.  Before all this DAW sophistication, it would have been possible, but really complicated, to do that level of processing all using hardware and miles of patch cords.  And in the analog world, there is the issue of accumulating noise at every connection poiont.  Is the level of fine tuning mentioned in this thread essentially the DAW equivalent of what always went on in the biggest studios?  Or do DAWs enable levels of processing that were rarely attempted in the analog studios of the past?
 
And another question along those lines.  It has always been a truism that the best studios were distinguished by using the most expensive, exotic microphones.  While I don't question that the most expensive mics can capture subtleties that other mics might miss, I wonder if a closet full of really expensive mics is as important today as it used to be.

Jeff M.
Many drum VSTs have multi outs -
Each will typically have different fx added to them

OK.  Makes sense.  Thanks.
 
 
And another question along those lines.  It has always been a truism that the best studies were distinguished by using the most expensive, exotic microphones.  While I don't question that the most expensive mics can capture subtleties that other mics might miss, I wonder if a closet full of really expensive mics is as important today as it used to be.

It might make sense for a studio owner or big-time producer, but for Joe Homeslice, probably not. I'd bet a wooden nickel you could trip up a few folks using  A-B blind comparos against Chinese and east European knock-offs. Of course, there are mic modeling plug-ins, such as the one Antares produces.
Then again, those exotics are quite nice, and vanity has never been a sin in the music world...

cparmerlee
I don't use any external hardware synths.  Everything is a software synth.  I'm trying to get a handle on best practices for that set-up.
 
I am really struggling to understand why anybody would ever use what seems to be the more popular method, creating separate MIDI and audio tracks and then routing the MIDI into a synth, then the output of the synth into the audio track.  It seems about 1000 times easier to simply insert an instrument track.  As far as I can tell, this achieves the same result with much less clutter.  You can put all the same audio effects on that track and do all the same automation.  Where is there any benefit in doing it the long way around?
 
I have been doing some arrangements with as many as 9 or 10 instrument tracks and this has been working great.  Almost no drop-outs and the latency is not a big problem.  I have not seen any need to freeze tracks.
 
I tried it the other way this afternoon (splitting into separate MIDI and audio tracks).  Not only was this far more cumbersome, it seemed to consume twice as much CPU power for the same exact voices, and this causes frequent drop-outs with most of the different synth modules.
 
All the posts I have read seem to imply that the "proper" way to do this is with split tracks.  My guess is that is the old way of doing things, and old habits die hard.  I can't see a single advantage to that method -- and am at the point of concluding it isn't even viable on my computer hardware.
 
Is there any advantage that I am missing?  Why would anybody ever split tracks when they can simply insert an Instrument Track?

Let's say you're using a VST like Kontakt.  And let's say you create a single instrument track.  Now suppose you need 4 different instruments loaded into Kontakt, each on a separate MIDI channel.  So now you create 3 additional MIDI tracks and route them to Kontakt.  It works.  The problem is that now when you trigger the active track and play back that instrument on MIDI channel 1, all the other tracks are receiving MIDI data as their tracks will "light up".  You won't hear the other tracks, only the active track, but it's still not good monitoring practice to see signal where there is none.  This is the situation where you want to create a separate MIDI and audio track--because when you add more tracks, the signals will be discreet.  Since every software synth handles issues differently (some allow for multiple, simultaneous instruments, some don't, some allow for patch changes, some don't) it's best to decide on a case by case basis.  For those synths that are only triggering one instrument at a time, then it makes sense to use a single instrument track.
 
JG
www.jerrygerber.com
 

bitflipper
The Simple Instrument Track works fine if you have a single non-multi-timbral synthesizer being driven by a single MIDI track. And you don't mind losing a few of the controls.
 
But if you're seriously into software synthesizers it's only a matter of time before you'll run into the limitations of the SIT. You may want to overlay two patches, slaving a second synth off the same MIDI track as the first. Can't do that with a SIT. You may want to use a multi-timbral sampler such as Kontakt for convenience and better CPU efficiency. Can't do that with a SIT. 

Actually, you can (I know it works, I've done it), but you shouldn't.  Reason explained in my other post preceding this one.
 
JG
www.jerrygerber.com
 

There's a bug in grouping SITs. For example, load one instance of SI-Drum Kit, one instance of SI-Bass, two instances of SI-Electric Piano and record a take on each track. Then group the Solo buttons of the two electric piano tracks, solo them and start playback. Do you hear anything? In my case, no I don't.

I always use separate audio and midi tracks.

• I find it more logical. Audio is audio and MIDI is MIDI, my simple brain finds the simple instrument track harder to comprehend.  For example I never quite get what the volume and pan faders do on an SIT, do they turn down or pan synth using MIDI or alter the volume and pan on the audio track. I suppose i could easily find out but it just seems easier to use a method where its all clearly layed out.
• A lot of the time I use multi out and/or multi-timber synths. For example an instance of Kontakt may contain eight instruments, with 8 stereo outs, driven by eight midi tracks, An Instance of an hammond Organ VSTi (B4 II) with one audio track and three midi tracks (upper & lower manuals and pedels) or Superior drummer with one midi track and a multitude of audio outs.
• Track templates make the setting up of these scenarios easy.
• The "clutter" can be managed, X2's auto track zoom is a great feature. Track folders are another big help One solution is to hide the MIDI tracks in consul view and hide the audio

 
 

I have had a couple of problems with SIT's, regarding the staff view.
When I try to do some edits, X2A crashes - and not on any on particular Synth.
If I do the same thing the 'old' way, no crashes.
 
 
 
Garry

charmerlee - I don't know about your skill level playing your material, but my skill level on keys needs extra help - which is why I never use simple instrument tracks.

With separate midi / audio tracks I clone the midi track as often as I want, so that I have multiple take tracks. They all route to the same audio track. So I can do bits and pieces of my performance and get clean parts, but they sound seamless to the audio track.

You could do the same thing with Take Lanes I suppose, but I'm not an adopter of those yet. If not using Take Lanes, the approach I've mentioned above only requires one instance of the synth say for five take tracks. If you had five alternate take tracks with SITs you'd have five synths loaded, which takes up unnecessary CPU.